Learning unification-based natural language grammars

Practical text processing systems need wide covering grammars. When parsing unrestricted language, such grammars often fail to generate all of the sentences that humans would judge to be grammatical. This problem undermines successful parsing of the text and is known as undergeneration. There are two main ways of dealing with undergeneration: either by sentence correction, or by grammar correction. This thesis concentrates upon automatic grammar correction (or machine learning of grammar) as a solution to the problem of undergeneration. Broadly speaking, grammar correction approaches can be classified as being either datadriven, or model-based. Data-driven learners use data-intensive methods to acquire grammar. They typically use grammar formalisms unsuited to the needs of practical text processing and cannot guarantee that the resulting grammar is adequate for subsequent semantic interpretation. That is, data-driven learners acquire grammars that generate strings that humans would judge to be grammatically ill-formed (they overgenerate) and fail to assign linguistically plausible parses. Model-based learners are knowledge-intensive and are reliant for success upon the completeness of a model of grammaticality. But in practice, the model will be incomplete. Given that in this thesis we deal with undergeneration by learning, we hypothesise that the combined use of data-driven and model-based learning would allow data-driven learning to compensate for model-based learning’s incompleteness, whilst model-based learning would compensate for data-driven learning’s unsoundness. We describe a system that we have used to test the hypothesis empirically. The system combines datadriven and model-based learning to acquire unification-based grammars that are more suitable for practical text parsing. Using the Spoken English Corpus as data, and by quantitatively measuring undergeneration, overgeneration and parse plausibility, we show that this hypothesis is correct.